Neuromorphic Knowledge Graph Processing With Spike-Time Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for processing and storing knowledge graphs require extensive offline data processing and are not suitable for real-time, online learning and inference in industrial settings due to their computational complexity and reliance on external data processing servers.
Innovation Solution
The development of neuromorphic hardware and methods that represent nodes and relations in knowledge graphs using spike times and spike time differences, enabling online learning and inference on edge devices with low-power, scalable processing units, such as neural processing units or neuromorphic processors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If existing systems for processing and storing knowledge graphs are used, then data can be stored and processed, but extensive offline data processing is required and real-time online learning is not possible due to computational complexity
Solution Approach 1:
The patent replaces traditional mechanical computing systems with neuromorphic hardware that mimics biological neural networks. This substitution enables parallel processing of knowledge graph data through spike-based temporal encoding, achieving real-time processing speeds while reducing computational complexity through event-driven architecture that only processes relevant data changes.
Solution Approach 2:
The patent transforms the representation parameters of knowledge graph data by using spike times and spike time differences instead of traditional static vectors. This parameter change enables dynamic, time-dependent encoding of node and relation information, allowing the system to process and learn from knowledge graphs in real-time with reduced computational overhead.
2Loss of time
If existing knowledge graph processing systems are used, then data analytics can be performed, but external data processing servers are required causing increased latency
Solution Approach 1:
The patent implements self-service by enabling industrial devices to perform knowledge graph processing and learning locally using embedded neuromorphic processors. Devices can independently store, process, and learn from their own data without requiring external data processing servers, thereby eliminating transmission latency and enabling immediate real-time decision-making.
Solution Approach 2:
The patent segments the centralized data processing architecture into distributed local processing units embedded in individual industrial devices. Each device contains its own neuromorphic processor that can independently handle knowledge graph operations, dividing the system into autonomous segments that reduce dependency on external servers and minimize data transmission delays.
3Use of energy by moving object
If traditional processing units are used, then computation can be performed, but power consumption is high and scalability is limited
Solution Approach 1:
The patent employs periodic action through event-driven spike-based processing where computations are triggered only by relevant data events rather than continuous processing. This periodic, on-demand computation significantly reduces power consumption compared to traditional continuous processing while maintaining high productivity by focusing computational resources only when necessary.
Solution Approach 2:
The patent implements dynamic processing through neuromorphic hardware that adapts its computational activity based on input patterns and learning requirements. The system dynamically adjusts its processing state, transitioning between active computation and low-power states, enabling energy-efficient operation that scales with actual workload demands rather than fixed power consumption.
Data Source
AI summary
Provided is neuromorphic hardware for storing and/or processing a knowledge graph with first neurons, representing a first node in the knowledge graph by first spike times of the first neurons during a recurring time interval, with second neurons, representing a second node in the knowledge graph by second spike times of the second neurons during the recurring time interval, and wherein a relation between the first node and the second node is represented as the differences between the first spike times and the second spike times.